Somewhere in your neighborhood there is a baby who "sleeps through the night," and her parents will tell you about it. What they won't tell you — because they don't know — is that their baby probably woke up last night. Possibly several times. She opened her eyes in the dark, maybe lifted her head, maybe chewed her fist for a minute. And then she went back to sleep without making a sound, and in the morning her parents woke up refreshed and concluded, reasonably but wrongly, that she had slept for eleven unbroken hours.

We know this because researchers have filmed babies overnight. And what the footage shows quietly dismantles the way most of us talk about infant sleep. The question that keeps exhausted parents up at 3 a.m. — when will my baby finally sleep through the night? — turns out to be the wrong question. Almost no baby sleeps through the night, not in the literal sense. The babies we call good sleepers are not sleeping more. They are waking, just like yours, and doing something different with the waking.

Nobody agreed on what "through the night" means

Start with an awkward fact: the phrase itself has no fixed definition. In one of the earliest systematic studies of infant sleep, back in the 1950s, researchers Moore and Ucko counted a baby as sleeping through if she stayed quiet from roughly midnight to five in the morning — five hours that conveniently overlap with when parents most want to be unconscious. Later researchers used a five-hour stretch anywhere in the night. Others argued that five hours is cold comfort for a parent whose baby is up at 4 a.m., and proposed stricter, family-friendly criteria closer to a full evening-to-morning stretch.

Which definition you pick changes the answer dramatically. Under the loose five-hour definition, many babies qualify within the first several months. Under the stay-asleep-as-long-as-the-parents definition, plenty of perfectly healthy babies don't get there in the first year. So when another parent says their four-month-old sleeps through, you are often comparing your strict definition against their generous one — and losing a competition that doesn't exist.

But the definitional mess hides a deeper problem. All of these criteria were based on what parents reported. And parents can only report the wakings they hear.

What the cameras saw

In the 1970s, the psychiatrist and sleep researcher Thomas Anders began doing something disarmingly simple: instead of asking parents how their babies slept, he set up time-lapse video and recorded the babies themselves, first in the lab and later in their own cribs at home. This technique, videosomnography, let researchers see the night as it actually happened rather than as it was remembered through a fog of exhaustion.

The recordings showed that waking at night is not the exception in the first year — it is the rule. Babies at every age studied surfaced from sleep during the night, often more than once. This makes biological sense. Infant sleep runs in short cycles, on the order of an hour or less, and the seam between one cycle and the next is a natural point of brief arousal. Adults have these arousals too; we roll over, adjust the pillow, and forget it ever happened. Babies are simply newer at the forgetting.

The difference that mattered was what happened after the waking. Anders and his colleagues found they could sort infants into two groups. Some babies woke and cried out — the researchers called them signalers. Others woke, looked around, perhaps fussed briefly or found a thumb, and returned to sleep on their own — the self-soothers. From the parents' bedroom, a self-soother is indistinguishable from a baby who never woke at all. "Sleeping through the night," the footage revealed, is largely a story parents tell about wakings they never heard.

This reframe is worth sitting with, because it changes what you're actually waiting for. Your baby is not going to develop the ability to sleep eleven hours in one seamless block, because human beings don't sleep that way at any age. What develops — partly on its own, partly with help — is the ability to cross the seam between sleep cycles without needing to summon you.

What has to mature, and what can be practiced

Some of the machinery of resettling is purely developmental, and no technique can rush it. Over the first months, sleep consolidates: the longest sleep period stretches out and migrates into the night as the circadian system comes online and the brain gets better at stringing cycles together. Tiny stomachs grow, and genuine nighttime hunger recedes on its own schedule. A newborn who wakes at 2 a.m. is usually waking because her body needs something. You cannot practice your way around that, and you shouldn't try.

But the videosomnography work pointed at something else, too: resettling has a learned component, and the learning hinges on what the moment of falling asleep looks like. A baby who drifts off under one set of conditions — being rocked, being fed, a parent's arms — and then surfaces at 2 a.m. under a completely different set of conditions wakes into a discrepancy. Something that was there is now missing, and the missing thing feels like the problem. A baby whose falling-asleep conditions are roughly the same as her middle-of-the-night conditions wakes into a familiar scene, and familiar scenes are easier to sleep through.

None of this means you must never rock or feed your baby to sleep — that's a separate decision with its own trade-offs, and warmth is not a mistake. It means the path from signaler to self-soother usually runs through two things: giving the baby occasional unassisted reps at the start of sleep, and not converting every brief arousal into a full waking by rushing in during the grunting, squirming, eyes-still-closed stage that often resolves on its own.

Timing matters too, in a way that's easy to miss. A baby who goes down overtired falls asleep on a surge of stress hormones, and stress hormones are the enemy of smooth cycle-to-cycle transitions. The best resettling practice happens in a baby who went to bed at the right moment — tired enough that sleep pressure carries her under, not so tired that her body is fighting itself all night.

Your next moves

  • Redefine the goal tonight. Stop waiting for zero wakings and start watching for silent resettles. If you use a monitor, note one waking this week where your baby stirred and returned to sleep alone — that's the skill growing, and it counts.
  • Give it sixty seconds. When you hear stirring at night, wait a full minute before going in (assuming no distress). Many arousals between sleep cycles resolve on their own if nobody escalates them into a wake-up.
  • Match the scenery. Make the conditions at bedtime resemble the conditions at 2 a.m.: same room, same darkness, same sound. If white noise will be on all night, have it on at sleep onset too, so nothing has vanished when she surfaces.
  • Practice one unassisted onset per day. Pick the easiest sleep of the day — often the first morning nap, when sleep pressure is strong — and put your baby down awake. One low-stakes rep a day builds the resettling muscle without turning bedtime into a battle.
  • Protect the bedtime window. An overtired baby resettles badly. Track today's wake windows and aim to start bedtime before the second-wind meltdown, not after it.

That last item is the quiet keystone. Resettling is a skill babies practice best from a calm start, and a calm start is mostly a matter of timing — catching the window when sleep pressure is high but the stress response hasn't kicked in. That window moves as your baby grows, which is exactly the moving target Drowsy was built to track. It learns your baby's rhythm and tells you the next realistic nap and bedtime window, so the reps that teach self-settling happen at the moment they're most likely to work. The wakings will keep coming — they come for every baby, the cameras proved it. Drowsy just helps you set up the nights where nobody needs to hear them.